OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
Image & Video AI

Ghost

GHOST (Generative High-fidelity One Shot Transfer) is an open-source Python research project that provides a one-shot face-swapping pipeline for images and video, with documented inference and model-training workflows.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Ghost — OpenViST project profile in Image & Video AI.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use GHOST when you need a research-oriented codebase and documented workflows for image or video face transfer, rather than a general-purpose editing service.

Who it is for

  • Computer-vision researchers exploring face-swapping methods
  • Developers who want to work with a documented Python inference and training repository
  • Educators creating hands-on demonstrations of face transfer

Documented uses

  • Swap a source face into a still image, with an optional specified target face.
  • Swap a source face into a video using target-face crops from the video.
  • Run a multi-person video swap by pairing multiple source faces with corresponding target faces.
  • Prepare VGGFace2 data and start training the face-swap model.
Before you choose

Requirements & limits

The repository is public and identifies the project as Apache-2.0 licensed. The README links Google Colab demos and a local Jupyter inference notebook.

  • Video workflows require target-face crops from the target video, and multi-person swaps require corresponding sets of faces.
  • The README describes local dependencies and model-weight downloads, so inference setup involves more than opening a hosted editing page.
  • The project reports that face-swap results remain imperfect in human and visual evaluation.
  • The project’s ethics statement rejects non-consensual or illicit use.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Set up GHOST for local inference

Prepare the repository, dependencies and model weights for the documented inference workflows.

Before you start

  • Python environment capable of installing the repository's requirements
  • Git
  • Ability to download the project’s model weights

Steps

  1. Clone the repository and initialize and update its submodules.

  2. Install dependencies from requirements.txt; the README suggests trying onnxruntime if onnxruntime-gpu cannot be installed.

  3. Download model weights by running the repository's download_models.sh script.

  4. Continue with the provided SberSwapInference.ipynb notebook or a documented inference command.

Expected outcome

The repository, dependencies and weights are prepared for the documented notebook or command-line inference stages.

Guide limitations

  • The README does not specify hardware requirements or guarantee compatibility across environments.
  • Video inference requires source image inputs and target-face crops; multi-person swaps require corresponding source and target faces.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
15 October 2026
Original website title
ai-forever/ghost · Official homepage title · retained in its source language
How we review sourcesSuggest a correction